Mechanisms of the Placebo Effect in Pain and Psychiatric Disorders

REVIEW Mechanisms of the placebo effect in pain and psychiatric disorders

RD Holmes, AK Tiwari and JL Kennedy

Placebo effect research over the past 15 years has improved our understanding of how placebo treatments reduce patient symptoms. The expectation of symptom improvement is the primary factor underlying the placebo effect. Such expectations are shaped by past experiences, contextual cues and biological traits, which ultimately modulate one’s degree of response to a placebo. The body of evidence that describes the physiology of the placebo effect has been derived from mechanistic studies primarily restricted to the setting of pain. Imaging ﬁndings support the role of endogenous opioid and dopaminergic networks in placebo analgesia in both healthy patients as well as patients with painful medical conditions. In patients with psychiatric illnesses such as anxiety disorders or depression, a vast overlap in neurological changes is observed in drug responders and placebo responders supporting the role of serotonergic networks in placebo response. Molecular techniques have been relatively underutilized in understanding the placebo effect until recently. We present an overview of the placebo responder phenotypes and genetic markers that have been associated with the placebo effect in pain, schizophrenia, anxiety disorders and depression.

The Pharmacogenomics Journal (2016) 16, 491–500; doi:10.1038/tpj.2016.15; published online 22 March 2016

INTRODUCTION lack of a study of familial inheritance. With genotyping assays and A tremendous amount of research has moved us closer to an whole gene sequencing increasingly being incorporated into the understanding of how placebo treatments can alleviate patient standard of care, pharmacogenomics is altering the approach to 11 symptoms in ways almost indistinguishable from pharmacother- prescribing medications. In this review, we will describe the apy. The earliest reports of placebo usage in the scientific characteristics of the placebo responder phenotype and current literature were mainly in the setting of painful medical hypotheses of placebo mechanisms with emphasis on recent conditions1 and date back at least to the 19th century.2 A seminal genetic studies in pain, schizophrenia, anxiety and depression. review by Beecher describes placebo treatments that reduced post-operative wound pain in 21–39% of individuals.3 In the past decade, numerous studies have reported placebo response rates THE PLACEBO EFFECT AND PAIN in excess of 40% and this trend of increasing placebo response has Placebo responder phenotype characteristics 4 been felt most prominently by pharmaceutical companies. Attempts to characterize the associated phenotypic traits of Success in clinical trials has decreased significantly in the past responders to placebo analgesia have been largely unsuccessful. A decade due to an inability to demonstrate therapeutic efficacy summary of sociodemographic and disease trait variables that are over a placebo control arm. For example, in schizophrenia, the predictive of placebo response in painful medical conditions is placebo response rate has been reported as the most influential provided in Table 1. The only trait found consistently among 5 predictor of trial success, contributing to the recent 15% increase placebo responders is a high baseline pain severity.12–14 Pain in phase 2 trial failures.6 These reports have piqued the interest of characteristics (for example, acute vs chronic) have not been many to investigate the underpinnings of the placebo effect found to affect placebo response.15 Only a few studies have using conventional technologies, such as electroencephalogram, reported variations in placebo response based on the age of positron emission tomography (PET), functional magnetic patients16,17 or level of education;17 however, a comparable resonance imaging and most recently, DNA microarrays. number of studies find no such associations.13,14 The incorporation of medical imaging into studies of the Beyond sociodemographics, certain behavioral and personality placebo effect has considerably expanded our understanding of traits have been associated with responders to placebo analgesia its neural mechanisms.7,8 Although molecular techniques have (see summary in Table 2). In an early study, placebo responders infiltrated most areas of biomedical research, they are markedly were more likely to be ‘talkers’, regular churchgoers, anxious, self- lacking in the study of the placebo effect. A small number of centered and describe their hospital care as ‘wonderful’, while investigations in search of serum biomarkers and metabolites non-responders were more ‘emotionally controlled’, ‘withdrawn’ provide a different perspective of the placebo effect; however, and ‘less comforted by the care received’.17 Consistent with their results remain challenging to incorporate into current placebo responders being described as ‘talkers’, extroversion has models.9,10 Unsurprisingly, the genetic underpinnings of the been a reported characteristic of placebo responders in the placebo effect are poorly characterized as demonstrated by the setting of irritable bowel syndrome (IBS)18 and experimentally

Neurogenetics Section, Neuroscience Department, Centre for Addiction and Mental Health, Toronto, Ontario, Canada. Correspondence: Professor JL Kennedy, Neurogenetics Section, Neuroscience Department, Centre for Addiction and Mental Health, 250 College Street, Toronto, Ontario M5T 1R8, Canada. E-mail: [email protected] Received 30 July 2014; revised 17 November 2015; accepted 20 January 2016; published online 22 March 2016 Mechanisms of the placebo effect RD Holmes et al 492

Table 1. Predictive variables of placebo phenotype in painful medical Table 2. Summary of personality traits predictive of placebo conditions responders in studies of experimentally induced pain as well as painful medical conditions Characteristic Condition References Characteristic Condition References Age (older) Wound pain Lasagna et al.17 16 Age (younger) Migraine Loder et al. High NEO altruism Experimental paina Peciña et al.23 17 Education (fewer years) Wound pain Lasagna et al. High NEO straight-forwardness Experimental paina Peciña et al.23 13 Baseline severity Neuropathic pain Irizarry et al. Low NEO angry hostility Experimental paina Peciña et al.23 12 (higher) (general) Zhang et al. High ego-resiliency Experimental paina Peciña et al.23 14 Osteoarthritis Häuser et al. Religiosity Wound pain Lasagna et al.17 14 Fibromyalgia syndrome Häuser et al. Novelty seeking Experimental paina Schweinhardt Painful diabetic et al.22 peripheral neuropathy Behavioral drive Experimental paina Schweinhardt et al.22 Fun seeking Experimental paina Schweinhardt 22 19 et al. induced pain. Thus far, dispositional optimism has received the Openness IBS Kelley et al.18 greatest support for having a role in the placebo effect. Extraversion Wound pain Lasagna et al.17 Investigators maintain personality as a moderating variable rather Experimental painb Morton et al.19 19–21 than a predictor variable of placebo effects. Dispositional IBS Kelley et al.18 optimism alone is not predictive of placebo response but has Dispositional optimism Wound pain Lasagna et al.17 been shown to have a statistically significant interaction with Experimental painb Morton et al.19 situational cues that vary patient expectancy.20,21 It was also found Experimental painc Geers et al.21 to be the most significant predictor of placebo response 19 Abbreviations: IBS, irritable bowel syndrome; NEO, neuroticism, extrover- reproducibility. sion, openness. aHypertonic saline masseter pain. bLaser-evoked Guided by findings of dopamine involvement in placebo cutaneous pain. cCold pressor pain. analgesia, researchers have found novelty seeking, behavioral drive and fun seeking to be correlated with placebo responders in 22 experimentally induced pain. In the same study, gray matter 28,29 density in the right ventral striatum, a region-of-interest in responses. Further investigation of the conditioned response suggests that the conditioning method determines if the opioi- functional neuroimaging studies, was also found to correlate with 29,30 dopamine-related personality traits as well as to the placebo effect. dergic system mediates the analgesic effects. The physiology A recent PET imaging study of μ-opioid activity during placebo of the placebo effect is also supported by functional neuroimaging administration in subjects who also underwent a battery of studies that demonstrate a correlation between a blood-oxygen level dependent signal in pain-responsive brain regions and the personality tests found placebo responders to have higher scores magnitude of placebo analgesia. The regions with the strongest in neuroticism, extroversion, openness (NEO) altruism, NEO straight- correlations include the rostral anterior cingulate cortex, contral- forwardness and ego-resiliency while having decreased scores in ateral insula, 1o somatosensory cortex (S1) and thalamus.31,32 NEO angry hostility.23 These four personality traits accounted for Coinciding with decreased activation of pain-responsive regions, a 25% of the variance in the placebo response and 27% of μ-opioid large number of studies have found evidence that supports the activity in the nucleus accumbens (NAC), a key structure in reward recruitment of the descending pain modulatory system to provide processing. This is the first evidence linking personality traits with – placebo analgesia.31,33 38 Imaging studies that have probed neural physiological activity in placebo responders, supporting the activity during the anticipation of pain have uncovered the hypothesis that personality has a role in placebo response. Given prominent role of the dorsolateral prefrontal cortex (DLPFC) and the numerous links made between genetic polymorphisms and orbitofrontal cortex (OFC). A heightened blood-oxygen level- personality traits, future studies in this area would be strengthened dependent signal in these regions correlates with subsequent through the incorporation of genetic testing.24 signal depression in pain responsive areas,32,39 whereas DLPFC μ- Key studies examining the stability of the phenotype call into opioid activity as measured by PET imaging negatively correlates question whether situation-independent traits can be reliably with the magnitude of anticipation of pain relief.34 In Figure 1, associated with placebo responders. Studies with repeated significant regions identified in PET and functional magnetic placebo administration find evidence to both support and refute resonance imaging studies have been delineated. the reliability and consistency of placebo responders.25 Experi- More recently, placebo analgesia has been framed as a form of ments in placebo analgesia have found that an individual can reward anticipation response that acts to modulate one’s switch from being a responder to non-responder simply by experience of a particular stimulus. Imaging evidence has been changing the labeling on the placebo’s container, while still found to support the involvement of both opioidergic and verbally suggesting it to be effective.26 Individuals who responded dopaminergic neural systems in this model.22,40,41 Expanding on to one set of environmental cues had a high likelihood of these investigations, Hall et al.42 performed one of the most responding again in the exact same environment suggesting that methodical genetic investigations into the placebo effect to date. the placebo response is reliable.19 Given the widely held belief of From a sample of 262 participants with IBS, 112 individuals were the placebo effect as a spurious phenomenon, efforts should be genetically screened for two SNPs (rs4633 and rs4680). The made to expand this area of placebo research to better functional polymorphism rs4680 results in an amino-acid change characterize the predictability of placebo responses. from valine to methionine (position 108/158) that decreases the activity of catechol-o-methyl transferase (COMT). COMT is a Mechanisms and genetics of placebo analgesia dopamine catabolizing enzyme critical in the prefrontal cortices The neurobiology of the placebo effect has been best character- and the presence of the Met allele of rs4680 results in significantly ized in placebo analgesia studies. The first biochemical evidence decreased enzymatic activity.43 COMT activity has been suggested of the placebo effect demonstrated a role for μ-opioid receptors in to influence areas such as cognitive performance,44 personality non-conditioned27 as well as conditioned placebo analgesia traits45 and reward processing.46 Patients were randomized to one

Figure 1. Regions of interest replicated in studies of placebo analgesia. Adapted from Stewart et al.130 DLPFC, dorsolateral prefrontal cortex; NAC, nucleus accumbens; OFC, orbitofrontal cortex; PAG, periaqueductal gray; rACC, rostral anterior cingulate cortex; VLPFC, ventrolateral prefrontal cortex. of three groups: (1) wait-list group, (2) limited placebo group or (3) in the NAC bilaterally. Interestingly, the level of NAC dopamine augmented placebo group. The often neglected wait-list group activity positively correlated with subjective pain ratings and a provides a control for spontaneous remission that can be a mediation analysis provided evidence that this correlation was substantial contributor to the placebo response.47 The limited and modulated by Met carrier status. During placebo administration, augmented placebo groups served to modulate patient expecta- the DA activity in the NAC was significantly different between Met tions with the augmented placebo group receiving the optimal carriers and Val homozygotes; however, during this stage of the practitioner–patient relationship. Patients homozygous for the low experiment, subjective pain ratings did not correlate with NAC DA activity Met allele of COMT overall had greater symptom activity. Although this meant that a genotype effect of rs6265 on improvement based on changes in IBS-Severity Scoring System placebo response could not be realized. This study demonstrated (IBS-SSS) score. More importantly, a statistically significant gene– a measurable difference between BDNF genotypes in the quantity environment interaction was found; patients with homozygous of NAC D2/3 receptors, in the level of DA activity in response to Met allele experienced significant symptom improvement with pain and in response to anticipated pain relief. These important enhanced patient–practitioner interaction while Val/Val genotype findings argue in favor of the hypothesis that the Met allele experienced no statistical effect in different therapeutic confers lower DA tone as well as lower DA response to reward. environments. Without concomitant imaging with a μ-opioid radiotracer, it is not Although a variation in COMT lends further support for the clear why in this instance a difference in placebo effect was not importance of the role of dopamine in placebo analgesia, the way observed. in which rs4680 complements imaging studies is not immediately A more recent study by Peciña et al. investigated the influence clear. Dopamine (DA) activity in reward processing structures is of the SNP rs1799971 (A118G; asparagine (Asn) to aspartic acid mainly regulated by dopamine transporters, not COMT. However, (Asp), Asn40Asp) in OPRM1 on placebo analgesia.53 Studies COMT activity has its greatest effect on dopamine in prefrontal suggest that G carriers express fewer μ-opioid receptors, which regions, such as the DLPFC, VLPFC and OFC, which are frequently results in decreased antinociceptive response to exogenous identified in functional magnetic resonance imaging studies of the opioids.55,56 A sample of 50 healthy volunteers underwent PET placebo effect. Investigators have found that one’s COMT imaging with [11C]raclopride and [11C]carfentanil at baseline, genotype explains a significant amount of DLPFC and limbic during a somatic pain challenge and during somatic pain with the system blood-oxygen level-dependent signals, with the Met COMT administration of a placebo. Expectations were modulated with allele associated with greater activation in response to unpleasant standardized statements that lead participants to believe that the visual stimuli.48 The DLPFC has been described as the entry point placebo was actually a potential medication capable of relieving of motivated behavior via its activation in anticipation of reward, pain. Consistent with the hypothesis, OPRM1 G carriers had lower which is consistent with DLPFC activation during anticipation of expression of μ-opioid receptors in regions previously implicated pain relief.32,39,49 In addition, attentional bias also has a reported in pain and emotional processing.53,57 However, in response to association with greater dopamine activity in the DLPFC.50 These pain without placebo, G carrier status was not correlated with pain results from literature support the finding of a potential role of threshold or opioid activity. In contrast, D2/3 receptor activity in COMT in the placebo effect. the NAC was significantly lower although it was uncorrelated with Further investigation into the impact of polymorphisms on the pain rating. Such a finding is inconsistent with the relationship interaction between dopaminergic and opioidergic systems has between pain perception and G carrier status identified by been undertaken with the common SNP rs6265 (Val66Met) in the others.58 Although examining placebo-induced neurotransmitter BDNF human gene and [11C]-raclopride PET imaging.51 It was changes, Montreal Pain Questionnaire (MPQ) sensory and affective postulated that Met carriers would have reduced activity- pain measures correlated with μ-opioid but not D2/3 activity in the dependent BDNF granule release,52 which could impact the NAC. This is supported by a similar study that demonstrated the dopaminergic neurons in reward-processing pathways.53,54 A same correlation with opioid activity.41 However, the current study small sample of 49 healthy volunteers (11Met carriers and 38 failed to demonstrate any correlation between pain severity and Val/Val) had expectations modulated by a standardized statement ACC opioid activity or NAC D2/3 activity, which have been leading them to believe that they would receive an experimental repeatedly identified.34,35,41 G carriers were found to have lower analgesic. During a somatic pain challenge before placebo opioid activity in the NAC, PAG, aINS, thalamus, regions repeatedly administration Met carriers were found to have greater DA activity identified as playing a significant role in placebo analgesia,

© 2016 Macmillan Publishers Limited, part of Springer Nature. The Pharmacogenomics Journal (2016), 491 – 500 Mechanisms of the placebo effect RD Holmes et al 494 although this did not translate into a significant separation in which is inline with the phenotypes of placebo analgesia and placebo responsiveness based on genotype. Methodologically, anxiolysis.72,73 Although estrogen is being studied as an adjunc- this study is a prototype for future studies seeking to disentangle tive medication in schizophrenia, this lack of differentiation of the relationship between the placebo response, neurologic placebo response based on sex suggests that it is less likely that activity and genetics. Sufficiently similar findings with the estrogen is an influential hormone in the neurobiology of placebo established literature support a possible role of OPRM1 A118G response in schizophrenia. in the placebo effect, although a more precise description of its There is evidence to suggest that the temporal profile of role will rely on future studies of this mutation as well as others in response can be a trait used to distinguish treatment responders key proteins of the opioidergic system. from placebo responders. In one retrospective analysis, Marques Beyond the dopamine–opioid interaction in placebo analgesia, et al.74 found that responses could be categorized in four ways: the endocannabinoid system has been found to impact the dramatic responders, responders, partial responders and non- placebo response in patients conditioned with non-opioid responders. Unfortunately, ~ 70% of patients in placebo or active analgesics.29 Endocannabinoids have been extensively studied treatment groups were responders or partial responders (that is, for their role in analgesia59 and reward processing60 and responded with the same temporal profile). The only significant specifically for their interaction with the opioidergic system.61 A difference was observed with dramatic responders; only patients recent genetic study of placebo analgesia incorporates PET receiving active treatment were dramatic responders. This result, imaging to specifically examine the impact of the rs324420 while based on only a small number of studies, suggests that (C385A) functional mutation that has a significant impact on placebo responders are more likely to respond late in the course endocannabinoid degradation.62 Specifically, rs324420 results in a of treatment. Although this result provides motivation for further Proline (Pro) to Threonine (Thr, position 129) missense substitution investigation others to date have not identified such a temporal 75 in fatty acid amide hydrolase reducing its catabolic activity of signature. endocannabinoids. In this study, 42 participants underwent a painful stimulus (that is, hypertonic saline in masseter) with and Neurobiology and genetic studies without a placebo (that is, IV normal saline) while being led to To the best of our knowledge, studies have not been conducted to believe that the therapeutic solution would have analgesic effects. investigate the neurobiology of the placebo effect in schizo- The participants were analyzed as two groups: Thr129 carrier or phrenia. However, based on the consistent trend that the placebo μ Pro/Pro. Both groups displayed statistically similar -opioid and effect works through disease-specific mechanisms,76 it is plausible D2/3 receptor activity for the same relative level of pain without a that the relief offered by placebo treatment in schizophrenia placebo. However, after a placebo was administered, Thr129 modifies the dopaminergic tone of mesolimbic and mesocortical carriers experienced a large placebo response in all metrics of the systems, which are responsible for the characteristic positive and Montreal Pain Questionnaire, which correlated with increases in μ- negative symptoms of schizophrenia. Evidence that the placebo opioid receptor activity in many structures identified by previous effect can recruit dopaminergic systems has been demonstrated placebo analgesia imaging studies (DLPFC, rostral anterior in other medical conditions such as pain and Parkinson’s disease, 34,41 cingulate cortex, thalamus, NAC and insula). The results of using PET imaging.41,77 In addition, COMT, a dopamine cataboliz- this study are very intriguing and contribute to the base of ing enzyme that exerts most of its influence in frontal regions, has evidence for a genetic component to the placebo effect. Due to been implicated in placebo responding patients with major the fact that endocannabinoid activity was not directly measured, depressive disorder78 and IBS.42 Future imaging genetic studies it is difficult to interpret why reduced analgesia was experienced focusing on key genes in dopaminergic functioning would provide 63 in Thr129 carriers, which is contrary to findings in animal studies. the foundation needed to understand the recent trend of In fact, fatty acid amide hydrolase knockout mice display CBR1- increasing placebo response. dependent pain reduction and increased levels of endocannabinoids.63,64 It is possible that the reconciliation of this discrepancy is due to fundamental differences in mouse and THE PLACEBO EFFECT AND ANXIETY human neuroanatomy. This study provides the foundation for Placebo responder phenotype characteristics future genetic-imaging studies that should attempt to elucidate The placebo response in the setting of anxiety strongly varies μ the dynamics of endocannabinoid and -opioid activity as it among anxiety disorder subtypes. Patients with obsessive relates to conditioned placebo analgesia. compulsive disorder have consistently been shown to have a small or even absent placebo response while studies of treatments 79,80 THE PLACEBO EFFECT AND SCHIZOPHRENIA for panic disorder have a notably high placebo response. Sociodemographic patient characteristics (for example, age, sex, Placebo responder phenotype characteristics social status, work status) have not been found to predict placebo The placebo response in schizophrenia is of growing concern as of response in panic disorder73,81 or social anxiety disorder.82 Similar late due to large clinical trial failures.6,65 Unfortunately, the to depression, as discussed later, the most reproduced patient majority of studies of placebo response in antipsychotic drug trials characteristic investigators have found to associate with placebo focus on identifying study characteristics related to an increased responders is lower initial illness severity.83–85 The temporal – response rate rather than placebo phenotype characteristics.66 69 pattern of placebo response appears to be early and sustained in A meta-analysis examining studies from 1970 to 2010 found placebo anxiolysis, which differs with reports of placebo response shorter trial duration to be associated with placebo response and patterns in depression.73,86 a follow-up meta-analysis replicated the observation of increased The study of the responder phenotype in placebo anxiolysis has placebo response rate in placebo-controlled randomized con- been somewhat complicated by discrepancies in placebo trolled trials (RCTs) since 1960.68,70 In addition, placebo response response related to outcome measures. Early investigations into has been found positively correlated with study sample size but differences between drug-treated and placebo-treated individuals not with the number of scheduled visits in RCTs.71 In terms of with panic disorder have found placebo anxiolysis to be effective patient or disease-related factors, younger age, greater baseline at decreasing somatic symptoms of anxiety (for example, panic severity and shorter illness duration have been associated with attacks), while less effective at influencing emotional aspects.87,88 increased placebo response.68 Two meta-analyses consistently The Hamilton Anxiety (HAM-A) questionnaire used to differentiate were unable to find sex to be predictive of placebo response, placebo response from medication response is biased toward the

The Pharmacogenomics Journal (2016), 491 – 500 © 2016 Macmillan Publishers Limited, part of Springer Nature. Mechanisms of the placebo effect RD Holmes et al 495 with increased activity in the OFC, ACC and VLPFC. In addition, Table 3. Predictive variables of placebo phenotype in depression self-reported relief positively correlated with activity in modula- Characteristic References N tory regions (that is, ACC and VLPFC) and negatively correlated with activity in perceptory regions (that is, extrastriate cortex and Sociodemographic characteristics amygdala). Similar changes in amygdala activity have been 99 15 Sex (female) Wilcox et al. 197 observed using [H2O ] PET imaging in placebo-treated patients Cohen et al.131 2533 105 with social anxiety disorder. Decreased regional cerebral blood Katon et al. 99 flow (rCBF) was observed in the left basomedial, left basolateral Marital status Wilcox et al.99 197 (married4widow4single) and right ventrolateral amygdala in those who responded to Age (older) Wilcox et al.99 197 placebo relative to those who did not respond. In addition, a Age (younger) Entsuah and 1651 comparison of rCBF changes between SSRI and placebo respon- Vinall98 ders found their neurophysiological changes to be indistingui- Race (non-Caucasians) Cohen et al.131 2533 shable.91 A follow-up analysis examining fronto-amygdala cou- NEO neuroticism (lower score) Katon et al.105 99 pling found remarkably similar co-activation patterns in placebo Education (fewer years) Katon et al.105 99 102 responders and SSRI responders, with increased DLPFC and rostral Information-processing speed (higher) Leuchter et al. 26 anterior cingulate cortex activation correlating with negative Characteristics of depression amygdala activity. This result lends support to the hypothesis that Baseline severity: HAMD (lower) Khan et al.132 2736 the neurobiological mechanisms of the placebo effect overlap 92 Stein et al.133 738 with those of active medications. Papakostas and 13 107 A substantial contribution to our understanding of the Fava97 molecular basis of placebo anxiolysis has been made by 15 82 Entsuah and 1651 combining genotyping with [H2O ] PET imaging. A relatively Vinall98 104 small sample of 24 patients with social anxiety disorder from two Brown et al. 241 RCTs underwent genotyping of the serotonin transporter length Wilcox et al.99 197 Bialik et al.109 99 polymorphism region (5-HTTLPR) and tryptophan hydroxylase-2 Katon et al.105 96 (TPH2) G-703T polymorphism (rs4570625). Both of these genes are Cohen et al.131 2533 pivotal in the proper functioning of the serotonergic system and ‘Late insomnia’ item score of HAMD Leuchter et al.102 26 have polymorphisms that result in a hyperexcitable amygdala. (higher) Specifically, individuals with the 5-HTTLPR s/s genotype and TPH2 Baseline severity: CGI-severity (lower) Cohen et al.131 2533 T-allele carriers display greater amygdala responsivity compared Depression subtype (reactive) Brown et al.104 241 93,94 104 with other genotypes under similarly stressful situations. The Prior psychiatric treatment (less) Brown et al. 241 investigators found that decreased rCBF to the left amygdala was Prior use of antidepressants (none) Brown et al.104 241 104 associated with placebo responders and that participants with the Experience with prior treatment Brown et al. 241 fi (successful) 5-HTTLPR l/l genotype displayed signi cantly reduced rCBF to the Illness duration (shorter) Fairchild et al.100 55 amygdala compared with the l/s or s/s genotypes (P = 0.004). As Brown et al.104 241 well, individuals with the TPH2 G/G genotype had significantly Cohen et al.131 2533 reduced rCBF to the amygdala compared with the G/T or T/T Illness duration (longer) Wilcox et al.99 197 alleles and as a result, dominated the group of placebo Non-endogenous symptoms (present) Fairchild et al.100 55 100 responders. In addition, individuals with both l/l 5-HTTLPR and History of abusive disorders (positive) Fairchild et al. 55 G/G TPH2 genotypes displayed the strongest placebo response. Abbreviations: CGI, Clinical Global Impression; HAMD, Hamilton Depression Surprisingly, rCBF in regions typically modulated in placebo Scale; NEO, neuroticism, extroversion, openness. N refers to the number of analgesia, namely the ACC, OFC, DLPFC and VLPFC, did not patients in the placebo group of each study.130 display any difference in rCBF between placebo responders and non-responders. Although these findings are an excellent step in the direction of measurement of somatic expressions of anxiety and under- uncovering the genetic underpinnings of the placebo effect, there emphasizes the emotional aspects, thus leading to seemingly are barriers to the full adoption of these results. Multiple factors in indistinguishable results between placebo and drug treatments. RCT design have been shown to influence placebo response rates 67,95,96 85 However, alternative outcome measures, such as Quality of Life in other diseases as well as anxiety disorders. When Enjoyment and Satisfaction, provide statistically significant differ- merging the data from multiple RCTs, it is very difficult to ensure similar environmental cues are held constant across the entire ences between placebo and drug treatment owing to their 82 broader scope.88 A similar issue has been described with using the sample. For example, the two RCTs used by Furmark et al. had percentage of patients that are panic free as an outcome measure different probabilities of randomization to placebo control. A link as it is uninfluenced by residual anticipatory anxiety and phobic between higher probability of randomization to placebo and 87 lower placebo response is thought to occur due to decreased avoidance. Therefore, future works in placebo anxioloysis should 97 be specifically cognizant of the outcome measures used to classify expectation of improvement. Future studies should take differences between RCTs into account when analyzing the response while trying to identify specific phenotypic traits. placebo effect. Mechanisms and genetics of placebo anxiolysis Brain imaging studies of individuals undergoing placebo anxiolysis THE PLACEBO EFFECT AND DEPRESSION are relatively lacking. The underlying source of anxiety has been Placebo responder phenotype characteristics traced to the hyper-responsivity of the amygdala and response to The placebo responder phenotype in depression has been treatment is often, but not always, associated with a decrease in extensively studied in the effort to reduce placebo response in responsivity.89 Petrovic et al.90 purport that placebo anxiolysis, like clinical trials. A summary of sociodemographic and disease traits placebo analgesia, is an example of reward processing with that have been studied for their predictive ability of placebo expectations of reduced anxiety in placebo responders associating response is provided in Table 3, and unfortunately few traits have

© 2016 Macmillan Publishers Limited, part of Springer Nature. The Pharmacogenomics Journal (2016), 491 – 500 Mechanisms of the placebo effect RD Holmes et al 496 strong evidence. Three reports could be found supporting a Although efforts to identify phenotypic traits of placebo correlation between age and placebo response. In a study-level responders have been relatively unyielding, the development of meta-analysis of 18 double-blind, placebo-controlled trials of the sequential parallel comparison design has led to a promising venlafaxine, age o40 years was associated with a significantly practical solution to reduce placebo response as well as required higher placebo response.98 Closer examination of these studies sample size.115 The study design involves the enrichment of reveals that patients under 18 were not included, and patients placebo non-responders from the first stage of the study, who older than 65 years were either not included or poorly then continue into the second stage. Drug efficacy is calculated represented. Wilcox et al.99 performed a patient-level meta- based on both stages of the study; a recent implementation of analysis of four studies that found a trend toward increased sequential parallel comparison design demonstrate its strength in placebo response among those 460 years old, although patients decreasing placebo response.116 over 60 made up 7.1% of the sample. Studies with evidence that contradicts this association were either underpowered or did not Mechanisms and genetics of placebo in depression compare placebo response at various age bins.100–103 Brown et al.104 performed the most balanced investigation of age in a Consistent with the placebo response in pain and anxiety, pooled analysis of three studies to yield no variation in placebo prefrontal regions have a prominent role in the resolution of response rate with age. MDD with placebo treatment. In an early study, prefrontal At this time, the literature does not support an association cordance, an electroencephalogram-derived biomarker of cortical blood flow, was seen to increase in placebo responders compared between gender and placebo response. Two small studies found 117 females to be more responsive to placebos and a large meta- to non-responders. Other investigations have found placebo analysis of children and adolescents including 27 RCTs found a responders and medications responders share the same pattern of trend toward the positive correlation between placebo response increased activation in the prefrontal cortex, posterior insula and and proportion females in the studies.99,105 An equally strong posterior cingulate, while displaying decreased activation in the fi 98,102–104 subgenual cingulate, parahippocampus and thalamus as mea- group of studies do not nd any association with gender. 18 Other traits such as marital status, education, race and NEO sured by ( Fluodeoxyglucose) PET imaging. The considerable neuroticism have also been studied although not enough studies similarity in alterations to glucose metabolism seen in both exist to warrant interpretation of these results. placebo responders and medication responders argues strongly In terms of major depressive disorder (MDD) characteristics for a shared pathway involving serotonin. among placebo responders, the most notable trend in the placebo Evidence for a role of the serotonergic system was revealed in response rate is that patients with a lower baseline severity of recent investigations into the genetics of placebo response, which depression are more likely to respond to placebo compared with examined candidate genes in the monoaminergic systems and 106 fi HPA axis for response to placebo (n = 257) as well as bupropion more severely depressed patients. This nding has been 118 consistently reproduced and provides one of the strongest (n = 319) in patients diagnosed with MDD. Genetic associations predictive measures used in forecasting models of placebo with placebo response or remission were found in the serotonin response.107,108 However, active medication response is relatively transporter (5-HTT), monoamine oxidase A (MAO-A) and serotonin stable across levels of depression, which is why increasing the 2A receptor (HTR2A). The strongest genetic association with minimum Hamilton Depression Scale score has been recom- placebo response or remission was observed with the A-allele of mended as a means of improving clinical trial performance. At this SNP rs4251417 in 5-HTT. Although this genetic association has not time, only three studies of illness duration have an adequate been observed in any other genetic studies of placebo response, sample size to interpret results and unfortunately, no consensus the authors hypothesize that the relevance of this SNP could be 82 exists based on length of current depressive episode or number of linked with the 5-HTTLPR. Other important SNPs that displayed previous episodes.98,104,109 Other MDD-specific traits such as nominal association with placebo response or remission include associated sleep disturbance or anxiety have been studied rs6609257, which is 6.6 kb downstream of MAO-A gene and although the body of literature is too nascent at this time to rs2296972 in HTR2A. The functional significance of rs6609257 in base judgements. MAO-A regulation is currently unknown; however, the results of Perhaps a lack of associable characteristics among placebo other studies suggest an important role of MAO-A regulation via 78 responders is due to the reliability and/or consistency of the the minor allele of rs6323 and the placebo response. MAO-A response. It seems logical that if placebo responders reliably involvement is further supported by the nominal association of respond to a placebo, they could be identified during the single placebo response with rs2235186, which is in complete linkage blind run-in period of RCTs and simply excluded, thus decreasing disequilibrium with rs6323. MAO-A modulates serotonergic tone 119 the placebo response rate and increasing drug–placebo differ- and serotonin has a demonstrated role in the reward pathway. ences. A meta-analysis of 42 trials (SSRIs vs placebo) comparing The last major finding relevant to the study of placebo was the trials that excluded placebo responders with those that did not association of HTR2A variant rs2296972 with placebo remission. exclude placebo responders, found no statistical difference in the This trend is important because it argues in favor of placebo effect size of treatment.110 In addition, earlier studies with similar effects working through mechanisms shared by the active objectives found that drug and placebo response rates were not medication of a treatment; SNPs in HTR2A have been associated affected by this practice.111,112 with response to bupropion118 and (es)citalopram120,121 among Investigators have also examined the temporal pattern in others. However, SNP rs2296972 that displays nominal association placebo responders with MDD as phenotypic trait. Katz et al.113 with placebo remission in this study is not described as having found that while the mean time to response was 13 days among association with antidepressants. It is possible that the HTR2A SSRI-treated patients, placebo-treated patients required 16 to SNPs associated with antidepressant response and placebo 42 days. This finding is supported by others that have described response share a functional variant although that has yet to be early symptom improvement to be twice as likely in antidepres- demonstrated. sant groups as placebo groups.114 A robust response to An earlier study of placebo response in the setting of MDD antidepressants has been associated with early symptom relief. specifically examined the Val158Met COMT polymorphism and the Such a claim also appears true for placebo treatment of MDD as Fnu4HI G/T polymorphism in MAO-A (rs6323) for association with successful statistical models have been developed to identify the placebo response in 52 patients.78 The investigators found placebo responders based on percentage change of HAMD-17 that homozygous patients or carriers of the rs6323 T-allele were score at week 2.107 more likely to respond to placebo (47% and 50%, respectively)

The Pharmacogenomics Journal (2016), 491 – 500 © 2016 Macmillan Publishers Limited, part of Springer Nature. Mechanisms of the placebo effect RD Holmes et al 497 than G-allele homozygotes. Patients with the rs6323 T-allele have RCTs, the path to including no-treatment control groups in future been associated with greater emotional reactivity to their studies is unclear. environment and greater interpersonal hypersensitivity.122,123 This Future genetic investigations would yield more impactful finding is congruous with the repeatedly demonstrated ability of findings by interrogating a larger number of genetic variants environmental cues to modulate patient expectations and the while concomitantly performing imaging. Although incorporating magnitude of placebo response. In the case of COMT, patients more genetic targets does exacerbate the issue of multiple with the high activity form of COMT (Val–Val) were found to have comparisons, such studies would provide a broader context with the highest placebo response. This finding lends support to the which to interpret findings. The combination of imaging with involvement of COMT in the placebo response; however, it is the genetics would ensure that future genetic studies do not yield 42 exact opposite genetic finding to that of placebo analgesia. insular findings in the overwhelmingly imaging-dominated field of Interestingly, individuals with the high activity form of COMT have placebo research. Improving on the weaknesses in current been found to have a stronger μ-opioid response to painful placebo genetics research highlighted in this review, genetics is stimuli124 and larger releases of dopamine in the nucleus fi 125 poised to add signi cant clarity to an intriguing yet enigmatic accumbens in response to a reward, making a case that phenomenon. individuals with high COMT activity could be more likely to respond to placebo. The role of endogenous opioid activity in depression is an active area of research, with recent studies CONFLICT OF INTEREST suggesting a role in the regulation of brain-derived neurotropic R.D.H and A.K.T have no conflicts of interest to declare. J.L.K. has been a consultant to factor, which is a central component to the pathology of GSK, Sanofi-Aventis and Dainippon-Sumitomo and received honoraria from Eli Lilly. depression.126,127 However, given our limited understanding of neural mechanisms in placebo treatment of depression, such speculations are at best tenuous. ACKNOWLEDGMENTS Beyond the theoretical limitations of not including a no- R.D.H. is supported by a CREMS Award 2013. A.K.T. is supported by the NARSAD 2010. treatment control, the methodological choice to combine Young Investigator Award. A.K.T. and J.L.K. are supported by the Ministry of Research different RCT samples reduces the sensitivity of these placebo and Innovation, Government of Ontario; and Genome Canada. genetics studies. A higher probability of receiving placebo has been correlated with a lower placebo response presumably due to REFERENCES lowered expectations of improvement.97,128,129 Leuchter et al.78 117 – used patients from four different clinical trials, one of which 1 Pepper OH. A note on the placebo. Am J Pharm 1945; :409 412. randomized 75% of participants to placebo, while the other three 2 Raicek JE, Stone BH, Kaptchuk TJ. Placebos in 19th century medicine: a quanti- tative analysis of the BMJ. BMJ 2012; 345:e8326. had a 50% chance of receiving a placebo. Similar heterogeneity is 159 – 118 3 Beecher H. The powerful placebo. J Am Med Assoc 1955; : 1602 1606. observed among the clinical trials included by Tiwari et al., 4 Kaptchuk TJ, Kelley JM, Conboy LA, Davis RB, Kerr CE, Jacobson EE et al. Com- although the use of a conditional logistic regression model ponents of placebo effect: randomised controlled trial in patients with irritable corrected for sociodemographic variables as well as the specific bowel syndrome. BMJ 2008; 336: 999–1003. trial would have partly accounted for the heterogeneity. Few 5 Khan A, Detke M, Khan SRF, Mallinckrodt C. Placebo response and anti- studies have investigated the role of genetic factors in placebo depressant clinical trial outcome. J Nerv Ment Dis 2003; 191:211–218. response in MDD and those that presently exist have been 6 Alphs L, Benedetti F, Fleischhacker WW, Kane JM. Placebo-related effects in restricted to traditional candidate genes. Genome-wide hypoth- clinical trials in schizophrenia: what is driving this phenomenon and what can be 15 – esis-free studies in larger samples are required to truly understand done to minimize it? Int J Neuropsychopharmacol 2012; : 1003 1014. 7 Faria V, Fredrikson M, Furmark T. Imaging the placebo response: a neurofunc- the role of genetic factors in placebo response. tional review. Eur Neuropsychopharmacol J Eur Coll Neuropsychopharmacol 2008; 18:473–485. RECOMMENDATIONS AND FUTURE DIRECTIONS 8 Lidstone SCC, Stoessl AJ. Understanding the placebo effect: contributions from neuroimaging. Mol Imaging Biol MIB Off Publ Acad Mol Imaging 2007; 9:176–185. Investigations into the placebo effect are growing in number and 9 Kokkotou E, Conboy LA, Ziogas DC, Quilty MT, Kelley JM, Davis RB et al. Serum with good reason; the placebo effect is a phenomenon with the correlates of the placebo effect in irritable bowel syndrome. Neurogastroenterol potential to provide extraordinary insight into the process of Motil 2010; 22:285–e81. healing. Althouggh research into the placebo effect has been 10 Kaddurah-Daouk R, Bogdanov MB, Wikoff WR, Zhu H, Boyle SH, Churchill E et al. studied extensively at the level of brain imaging, genetics and Pharmacometabolomic mapping of early biochemical changes induced by ser- other molecular methods (for example, metabolomics) have yet to traline and placebo. Transl Psychiatry 2013; 3: e223. be fully implemented in this field. In general, the genetic studies 11 Cordero P, Ashley EA. Whole-genome sequencing in personalized therapeutics. 91 – presented in this review provide complementary evidence as well Clin Pharmacol Ther 2012; : 1001 1009. 12 Zhang W, Robertson J, Jones AC, Dieppe PA, Doherty M. The placebo effect and as a new lens through which to scrutinize current models of the its determinants in osteoarthritis: meta-analysis of randomised controlled trials. placebo effect. The genetic-imaging studies have generally Ann Rheum Dis 2008; 67: 1716–1723. focused on analyzing single polymorphism from a gene of 13 Irizarry MC, Webb DJ, Ali Z, Chizh BA, Gold M, Kinrade FJ et al. Predictors of interest. In the future, it would be interesting to see how gene– placebo response in pooled lamotrigine neuropathic pain clinical trials. Clin J gene interaction (for example, OPRM1 and fatty acid amide Pain 2009; 25:469–476. hydrolase) contribute to placebo effect. 14 Häuser W, Bartram-Wunn E, Bartram C, Reinecke H, Tölle T. Systematic review: Moving forward improvements in study design are warranted to placebo response in drug trials of fibromyalgia syndrome and painful peripheral improve the sensitivity of genetic studies and to ensure results are diabetic neuropathy-magnitude and patient-related predictors. Pain 2011; 152: reproducible. Current genetic studies in general are limited due to 1709–1717. 42 15 Kamper SJ, Machado LAC, Herbert RD, Maher CG, McAuley JH. Trial methodology a lack of no-treatment control groups (except for Hall et al. ), fl using samples with heterogeneous therapeutic experiences, and and patient characteristics did not in uence the size of placebo effects on pain. J Clin Epidemiol 2008; 61:256–260. relatively small sample sizes. Without question the most important 16 Loder E, Goldstein R, Biondi D. Placebo effects in oral triptan trials: the scientific recommendation for future genetic investigations is the inclusion and ethical rationale for continued use of placebo controls. Cephalalgia Int J of a no-treatment control group as the interpretation of findings Headache 2005; 25:124–131. will always remain tenuous without such a measure. Given the 17 Lasagna L, Mosteller F, Von Felsinger JM, Beecher HK. A study of the placebo ethical dilemma surrounding the use of placebo-control groups in response. Am J Med 1954; 16:770–779.

© 2016 Macmillan Publishers Limited, part of Springer Nature. The Pharmacogenomics Journal (2016), 491 – 500 Mechanisms of the placebo effect RD Holmes et al 498 18 Kelley JM, Lembo AJ, Ablon JS, Villanueva JJ, Conboy LA, Levy R et al. Patient and 46 Yacubian J, Sommer T, Schroeder K, Gläscher J, Kalisch R, Leuenberger B et al. practitioner influences on the placebo effect in irritable bowel syndrome. Psy- Gene–gene interaction associated with neural reward sensitivity. Proc Natl Acad chosom Med 2009; 71: 789–797. Sci USA 2007; 104: 8125–8130. 19 Morton DL, Watson A, El-Deredy W, Jones AKP. Reproducibility of placebo 47 Krogsbøll LT, Hróbjartsson A, Gøtzsche PC. Spontaneous improvement in ran- analgesia: effect of dispositional optimism. Pain 2009; 146: 194–198. domised clinical trials: meta-analysis of three-armed trials comparing no treat- 20 Geers AL, Helfer SG, Kosbab K, Weiland PE, Landry SJ. Reconsidering the role of ment, placebo and active intervention. BMC Med Res Methodol 2009; 9:1. personality in placebo effects: dispositional optimism, situational expectations, 48 Smolka MN, Schumann G, Wrase J, Grüsser SM, Flor H, Mann K et al. Catechol-O- and the placebo response. J Psychosom Res 2005; 58:121–127. methyltransferase val158met genotype affects processing of emotional stimuli 21 Geers AL, Wellman JA, Fowler SL, Helfer SG, France CR. Dispositional in the amygdala and prefrontal cortex. J Neurosci Off J Soc Neurosci 2005; 25: optimism predicts placebo analgesia. J Pain Off J Am Pain Soc 2010; 11: 836–842. 1165–1171. 49 Ballard IC, Murty VP, Carter RM, MacInnes JJ, Huettel SA, Adcock RA. Dorsolateral 22 Schweinhardt P, Seminowicz DA, Jaeger E, Duncan GH, Bushnell MC. The anat- prefrontal cortex drives mesolimbic dopaminergic regions to initiate motivated omy of the mesolimbic reward system: a link between personality and the behavior. J Neurosci Off J Soc Neurosci 2011; 31: 10340–10346. placebo analgesic response. J Neurosci 2009; 29: 4882–4887. 50 Luijten M, Veltman DJ, Hester R, Smits M, Pepplinkhuizen L, Franken IHA. Brain 23 Peciña M, Azhar H, Love TM, Lu T, Fredrickson BL, Stohler CS et al. Personality activation associated with attentional bias in smokers is modulated by a trait predictors of placebo analgesia and neurobiological correlates. Neu- dopamine antagonist. Neuropsychopharmacology 2012; 37: 2772–2779. ropsychopharmacology 2013; 38:639–646. 51 Peciña M, Martínez-Jauand M, Love T, Heffernan J, Montoya P, Hodgkinson C 24 de Moor MHM, Costa PT, Terracciano A, Krueger RF, de Geus EJC, Toshiko T et al. et al. Valence-specific effects of BDNF Val66Met polymorphism on dopaminergic Meta-analysis of genome-wide association studies for personality. Mol Psychiatry stress and reward processing in humans. J Neurosci Off J Soc Neurosci 2014; 34: 17 – 2012; : 337 349. 5874–5881. 25 Kaptchuk TJ, Kelley JM, Deykin A, Wayne PM, Lasagna LC, Epstein IO et al. Do 52 Chen Z-Y, Patel PD, Sant G, Meng C-X, Teng KK, Hempstead BL et al. Variant ‘ ’ 29 – placebo responders exist? Contemp Clin Trials 2008; : 587 595. brain-derived neurotrophic factor (BDNF) (Met66) alters the intracellular traf- 26 Whalley B, Hyland ME, Kirsch I. Consistency of the placebo effect. J Psychosom ficking and activity-dependent secretion of wild-type BDNF in neurosecretory 64 – Res 2008; : 537 541. cells and cortical neurons. J Neurosci Off J Soc Neurosci 2004; 24: 4401–4411. 27 Levine JD, Gordon NC, Fields HL. The mechanism of placebo analgesia. Lancet 53 Peciña M, Love T, Stohler CS, Goldman D, Zubieta J-K. Effects of the Mu opioid 2 – 1978; : 654 657. receptor polymorphism (OPRM1 A118G) on pain regulation, placebo effects and 28 Benedetti F, Amanzio M, Maggi G. Potentiation of placebo analgesia by pro- associated personality trait measures. Neuropsychopharmacology 2015; 40: glumide. Lancet 1995; 346: 1231. 957–965. 29 Benedetti F, Amanzio M, Rosato R, Blanchard C. Nonopioid placebo analgesia is 54 Nees F, Witt SH, Dinu-Biringer R, Lourdusamy A, Tzschoppe J, Vollstädt-Klein S mediated by CB1 cannabinoid receptors. Nat Med 2011; 17: 1228–1230. et al. BDNF Val66Met and reward-related brain function in adolescents: role for 30 Amanzio M, Benedetti F. Neuropharmacological dissection of placebo analgesia: early alcohol consumption. Alcohol 2015; 49: 103–110. expectation-activated opioid systems versus conditioning-activated specific 55 Kroslak T, Laforge KS, Gianotti RJ, Ho A, Nielsen DA, Kreek MJ. The single subsystems. J Neurosci 1999; 19:484–494. nucleotide polymorphism A118G alters functional properties of the human mu 31 Eippert F, Bingel U, Schoell ED, Yacubian J, Klinger R, Lorenz J et al. Activation of opioid receptor. J Neurochem 2007; 103:77–87. the opioidergic descending pain control system underlies placebo analgesia. 56 Sia AT, Lim Y, Lim ECP, Goh RWC, Law HY, Landau R et al. A118G single Neuron 2009; 63:533–543. nucleotide polymorphism of human mu-opioid receptor gene influences pain 32 Wager TD, Rilling JK, Smith EE, Sokolik A, Casey KL, Davidson RJ et al. Placebo- perception and patient-controlled intravenous morphine consumption after induced changes in fMRI in the anticipation and experience of pain. Science intrathecal morphine for postcesarean analgesia. Anesthesiology 2008; 109: 2004; 303: 1162–1167. 520–526. 33 Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia-- 57 Zubieta JK, Smith YR, Bueller JA, Xu Y, Kilbourn MR, Jewett DM et al. Regional mu imaging a shared neuronal network. Science 2002; 295: 1737–1740. opioid receptor regulation of sensory and affective dimensions of pain. Science 34 Zubieta J-K, Bueller JA, Jackson LR, Scott DJ, Xu Y, Koeppe RA et al. Placebo 2001; 293: 311–315. effects mediated by endogenous opioid activity on μ-opioid receptors. J Neu- 58 Oertel BG, Schmidt R, Schneider A, Geisslinger G, Lötsch J. The mu-opioid rosci 2005; 25: 7754–7762. 4 35 Wager TD, Scott DJ, Zubieta J-K. Placebo effects on human mu-opioid activity receptor gene polymorphism 118A G depletes alfentanil-induced analgesia during pain. Proc Natl Acad Sci USA 2007; 104: 11056–11061. and protects against respiratory depression in homozygous carriers. Pharma- 16 – 36 Eippert F, Finsterbusch J, Bingel U, Büchel C. Direct evidence for spinal cord cogenet Genomics 2006; :625 636. involvement in placebo analgesia. Science 2009; 326:404–404. 59 Hohmann AG. Spinal and peripheral mechanisms of cannabinoid antinocicep- 37 Stein N, Sprenger C, Scholz J, Wiech K, Bingel U. White matter integrity of the tion: behavioral, neurophysiological and neuroanatomical perspectives. Chem 121 – descending pain modulatory system is associated with interindividual differ- Phys Lipids 2002; :173 190. ences in placebo analgesia. Pain 2012; 153: 2210–2217. 60 Gardner EL, Vorel SR. Cannabinoid transmission and reward-related events. 5 – 38 Colloca L, Klinger R, Flor H, Bingel U. Placebo analgesia: psychological and Neurobiol Dis 1998; :502 533. neurobiological mechanisms. Pain 2013; 154:511–514. 61 Welch SP. Interaction of the cannabinoid and opioid systems in the modulation 21 – 39 Wager TD, Atlas LY, Leotti LA, Rilling JK. Predicting individual differences in of nociception. Int Rev Psychiatry Abingdon Engl 2009; : 143 151. placebo analgesia: contributions of brain activity during anticipation and pain 62 Peciña M, Martínez-Jauand M, Hodgkinson C, Stohler CS, Goldman D, fl 19 experience. J Neurosci 2011; 31:439–452. Zubieta JK. FAAH selectively in uences placebo effects. Mol Psychiatry 2014; : 40 Scott DJ, Stohler CS, Egnatuk CM, Wang H, Koeppe RA, Zubieta J-K. Individual 385–391. differences in reward responding explain placebo-induced expectations and 63 Cravatt BF, Demarest K, Patricelli MP, Bracey MH, Giang DK, Martin BR et al. effects. Neuron 2007; 55:325–336. Supersensitivity to anandamide and enhanced endogenous cannabinoid sig- 41 Scott DJ, Stohler CS, Egnatuk CM, Wang H, Koeppe RA, Zubieta J-K. Placebo and naling in mice lacking fatty acid amide hydrolase. Proc Natl Acad Sci USA 2001; nocebo effects are defined by opposite opioid and dopaminergic responses. 98: 9371–9376. Arch Gen Psychiatry 2008; 65:220–231. 64 Clement AB, Hawkins EG, Lichtman AH, Cravatt BF. Increased seizure suscept- 42 Hall KT, Lembo AJ, Kirsch I, Ziogas DC, Douaiher J, Jensen KB et al. Catechol-O- ibility and proconvulsant activity of anandamide in mice lacking fatty acid amide methyltransferase val158met polymorphism predicts placebo effect in irritable hydrolase. J Neurosci Off J Soc Neurosci 2003; 23: 3916–3923. bowel syndrome. PloS One 2012; 7: e48135. 65 Kemp AS, Schooler NR, Kalali AH, Alphs L, Anand R, Awad G et al. What is causing 43 Meyer-Lindenberg A, Kohn PD, Kolachana B, Kippenhan S, McInerney-Leo A, the reduced drug-placebo difference in recent schizophrenia clinical trials and Nussbaum R et al. Midbrain dopamine and prefrontal function in what can be done about it? Schizophr Bull 2010; 36:504–509. humans: interaction and modulation by COMT genotype. Nat Neurosci 2005; 8: 66 Volavka J, Cooper TB, Laska EM, Meisner M. Placebo washout in trials of 594–596. antipsychotic drugs. Schizophr Bull 1996; 22:567–576. 44 Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE et al. 67 Mallinckrodt CH, Tamura RN, Tanaka Y. Recent developments in improving Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for signal detection and reducing placebo response in psychiatric clinical trials. schizophrenia. Proc Natl Acad Sci USA 2001; 98: 6917–6922. J Psychiatr Res 2011; 45: 1202–1207. 45 Dávila W, Basterreche N, Arrue A, Zamalloa MI, Gordo E, Dávila R et al. The 68 Agid O, Siu CO, Potkin SG, Kapur S, Watsky E, Vanderburg D et al. Meta- influence of the Val158Met catechol-O-methyltransferase polymorphism on the regression analysis of placebo response in antipsychotic trials, 1970-2010. Am J personality traits of bipolar patients. PloS One 2013; 8: e62900. Psychiatry 2013; 170: 1335–1344.

The Pharmacogenomics Journal (2016), 491 – 500 © 2016 Macmillan Publishers Limited, part of Springer Nature. Mechanisms of the placebo effect RD Holmes et al 499 69 Khan A, Yavorsky WC, Liechti S, DiClemente G, Rothman B, Opler M et al. 95 Potter WZ, Demitrack MA, DeBrota DJ, Faries D, Herrera J. Controlling the pla- Assessing the sources of unreliability (rater, subject, time-point) in a failed cebo response rates in depression. Eur Neuropsychopharmacol 1998; 8 (Sup- clinical trial using items of the Positive and Negative Syndrome Scale (PANSS). plement 2): S80–S81. J Clin Psychopharmacol 2013; 33: 109–117. 96 Patel SM, Stason WB, Legedza A, Ock SM, Kaptchuk TJ, Conboy L et al. The 70 Welge JA, Keck PE Jr. Moderators of placebo response to antipsychotic treat- placebo effect in irritable bowel syndrome trials: a meta-analysis. Neurogas- ment in patients with schizophrenia: a meta-regression. Psychopharmacology troenterol Motil 2005; 17:332–340. (Berl) 2003; 166:1–10. 97 Papakostas GI, Fava M. Does the probability of receiving placebo influence 71 Rutherford BR, Pott E, Tandler JM, Wall MM, Roose SP, Lieberman JA. Placebo clinical trial outcome? A meta-regression of double-blind, randomized clinical response in antipsychotic clinical trials: A meta-analysis. JAMA Psychiatry 2014; trials in MDD. Eur Neuropsychopharmacol 2009; 19:34–40. 71: 1409–1421. 98 Entsuah R, Vinall P. Potential predictors of placebo response: lessons from a large 72 Averbuch M, Katzper M. Gender and the placebo analgesic effect in acute pain. database. Drug Inf J 2007; 41: 315–330. Clin Pharmacol Ther 2001; 70:287–291. 99 Wilcox CS, Cohn JB, Linden RD, Heiser JF, Lucas PB, Morgan DL et al. Predictors of 73 Dager SR, Khan A, Cowley D, Avery DH, Elder J, Roy-Byrne P et al. Characteristics placebo response: a retrospective analysis. Psychopharmacol Bull 1992; 28: of placebo response during long-term treatment of panic disorder. Psycho- 157–162. pharmacol Bull 1990; 26:273–278. 100 Fairchild CJ, Rush AJ, Vasavada N, Giles DE, Khatami M. Which depressions 74 Marques TR, Arenovich T, Agid O, Sajeev G, Muthén B, Chen L et al. The different respond to placebo? Psychiatry Res 1986; 18:217–226. trajectories of antipsychotic response: antipsychotics versus placebo. Psychol 101 Sheline YI, Black KJ, Bardgett ME, Csernansky JG. Platelet binding characteristics Med 2011; 41: 1481–1488. distinguish placebo responders from nonresponders in depression. Neu- 75 Potkin S, Agid O, Siu C, Watsky E, Vanderburg D, Remington G. Placebo response ropsychopharmacology 1995; 12:315–322. trajectories in short-term and long-term antipsychotic trials in schizophrenia. 102 Leuchter AF, Morgan M, Cook IA, Dunkin J, Abrams M, Witte E. Pretreatment Schizophr Res 2011; 132: 108–113. neurophysiological and clinical characteristics of placebo responders in treat- 76 Benedetti F. Mechanisms of placebo and placebo-related effects across diseases ment trials for major depression. Psychopharmacology (Berl) 2004; 177:15–22. and treatments. Annu Rev Pharmacol Toxicol 2008; 48:33–60. 103 Brown WA, Dornseif BE, Wernicke JF. Placebo response in depression: a search 77 de la Fuente-Fernández R, Schulzer M, Stoessl AJ. Placebo mechanisms and for predictors. Psychiatry Res 1988; 26:259–264. reward circuitry: clues from Parkinson’s disease. Biol Psychiatry 2004; 56:67–71. 104 Brown WA, Johnson MF, Chen MG. Clinical features of depressed patients who 78 Leuchter AF, McCracken JT, Hunter AM, Cook IA, Alpert JE. Monoamine oxidase a do and do not improve with placebo. Psychiatry Res 1992; 41:203–214. and catechol-o-methyltransferase functional polymorphisms and the placebo 105 Katon W, Russo J, Frank E, Barrett J, Williams JW Jr, Oxman T et al. Predictors of response in major depressive disorder. J Clin Psychopharmacol 2009; 29: nonresponse to treatment in primary care patients with dysthymia. Gen Hosp 372–377. Psychiatry 2002; 24:20–27. 79 Mavissakalian MR, Jones B, Olson S. Absence of placebo response in obsessive- 106 Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, Johnson BT. Initial compulsive disorder. J Nerv Ment Dis 1990; 178:268–270. severity and antidepressant benefits: a meta-analysis of data submitted to the 80 Huppert JD, Schultz LT, Foa EB, Barlow DH, Davidson JRT, Gorman JM et al. food and drug administration. PLoS Med 2008; 5: e45. Differential response to placebo among patients with social phobia, panic dis- 107 Gomeni R, Merlo-Pich E. Bayesian modelling and ROC analysis to predict placebo order, and obsessive-compulsive disorder. Am J Psychiatry 2004; 161: 1485–1487. responders using clinical score measured in the initial weeks of treatment in 81 Rosenberg NK, Mellergård M, Rosenberg R, Beck P, Ottosson JO. Characteristics depression trials. Br J Clin Pharmacol 2007; 63:595–613. of panic disorder patients responding to placebo. Acta Psychiatr Scand Suppl 108 Nelson JC, Zhang Q, Deberdt W, Marangell LB, Karamustafalioglu O, Lipkovich IA. 1991; 365:33–38. Predictors of remission with placebo using an integrated study database 82 Furmark T, Appel L, Henningsson S, Ahs F, Faria V, Linnman C et al. A link from patients with major depressive disorder. Curr Med Res Opin 2012; 28: between serotonin-related gene polymorphisms, amygdala activity, and 325–334. placebo-induced relief from social anxiety. J Neurosci Off J Soc Neurosci 2008; 28: 109 Bialik RJ, Ravindran AV, Bakish D, Lapierre YD. A comparison of placebo 13066–13074. responders and nonresponders in subgroups of depressive disorder. J Psychiatry 83 Coryell W, Noyes R. Placebo response in panic disorder. Am J Psychiatry 1988; Neurosci 1995; 20:265–270. 145: 1138–1140. 110 Lee S, Walker JR, Jakul L, Sexton K. Does elimination of placebo responders in a 84 Woodman CL, Noyes R Jr, Ballenger JC, Lydiard RB, Sievers G, Mihalko D. Pre- placebo run-in increase the treatment effect in randomized clinical trials? dictors of response to alprazolam and placebo in patients with panic disorder. A meta-analytic evaluation. Depress Anxiety 2004; 19:10–19. J Affect Disord 1994; 30:5–13. 111 Trivedi MH, Rush H. Does a placebo run-in or a placebo treatment cell affect the 85 Hackett D, Haudiquet V, Salinas E. A method for controlling for a high placebo efficacy of antidepressant medications? Neuropsychopharmacolopgy 1994; 11: response rate in a comparison of venlafaxine XR and diazepam in the short-term 33–43. treatment of patients with generalised anxiety disorder. Eur Psychiatry J Assoc Eur 112 Greenberg RP, Fisher S, Riter JA. Placebo washout is not a meaningful part of Psychiatr 2003; 18: 182–187. antidepressant drug trials. Percept Mot Skills 1995; 81:688–690. 86 Schweizer E, Rickels K. Placebo response in generalized anxiety: its effect on the 113 Katz MM, Tekell JL, Bowden CL, Brannan S, Houston JP, Berman N et al. Onset outcome of clinical trials. J Clin Psychiatry 1997; 58 (Suppl 11): 30–38. and early behavioral effects of pharmacologically different antidepressants and 87 Rosenberg R. Prediction of placebo response in panic disorder: a short review. placebo in depression. Neuropsychopharmacology 2004; 29:566–579. Nord J Psychiatry 1994; 48:153–158. 114 Papakostas GI, Perlis RH, Scalia MJ, Petersen TJ, Fava M. A meta-analysis of early 88 Rapaport MH, Pollack M, Wolkow R, Mardekian J, Clary C. Is placebo sustained response rates between antidepressants and placebo for the treat- response the same as drug response in panic disorder? Am J Psychiatry 2000; ment of major depressive disorder. J Clin Psychopharmacol 2006; 26:56–60. 157: 1014–1016. 115 Fava M, Evins AE, Dorer DJ, Schoenfeld DA. The problem of the placebo response 89 Domschke K, Dannlowski U. Imaging genetics of anxiety disorders. NeuroImage in clinical trials for psychiatric disorders: Culprits, possible remedies, and a novel 2010; 53:822–831. study design approach. Psychother Psychosom 2003; 72:115–127. 90 Petrovic P, Dietrich T, Fransson P, Andersson J, Carlsson K, Ingvar M. Placebo in 116 Fava M, Mischoulon D, Iosifescu D, Witte J, Pencina M, Flynn M et al. A double- emotional processing--induced expectations of anxiety relief activate a blind, placebo-controlled study of aripiprazole adjunctive to antidepressant generalized modulatory network. Neuron 2005; 46: 957–969. therapy among depressed outpatients with inadequate response to prior anti- 91 Faria V, Appel L, Åhs F, Linnman C, Pissiota A, Frans Ö et al. Amygdala subregions depressant therapy (ADAPT-A Study). Psychother Psychosom 2012; 81:87–97. tied to SSRI and placebo response in patients with social anxiety disorder. 117 Leuchter AF, Cook IA, Witte EA, Morgan M, Abrams M. Changes in brain function Neuropsychopharmacology 2012; 37: 2222–2232. of depressed subjects during treatment with placebo. Am J Psychiatry 2002; 159: 92 Benedetti F, Carlino E, Pollo A. How placebos change the patient’s brain. Neu- 122–129. ropsychopharmacology 2011; 36: 339–354. 118 Tiwari AK, Zai CC, Sajeev G, Arenovich T, Müller DJ, Kennedy JL. Analysis of 34 93 Furmark T, Tillfors M, Garpenstrand H, Marteinsdottir I, Långström B, Oreland L candidate genes in bupropion and placebo remission. Int J Neuropsycho- et al. Serotonin transporter polymorphism related to amygdala excitability and pharmacol 2012: 1–11. symptom severity in patients with social phobia. Neurosci Lett 2004; 362: 119 Kranz GS, Kasper S, Lanzenberger R. Reward and the serotonergic system. 189–192. Neuroscience 2010; 166: 1023–1035. 94 Brown SM, Peet E, Manuck SB, Williamson DE, Dahl RE, Ferrell RE et al. A reg- 120 McMahon FJ, Buervenich S, Charney D, Lipsky R, Rush AJ, Wilson AF et al. Var- ulatory variant of the human tryptophan hydroxylase-2 gene biases amygdala iation in the gene encoding the serotonin 2A receptor is associated with out- reactivity. Mol Psychiatry 2005; 10:884–888. come of antidepressant treatment. Am J Hum Genet 2006; 78:804–814.

© 2016 Macmillan Publishers Limited, part of Springer Nature. The Pharmacogenomics Journal (2016), 491 – 500 Mechanisms of the placebo effect RD Holmes et al 500 121 Uher R, Huezo-Diaz P, Perroud N, Smith R, Rietschel M, Mors O et al. Genetic 128 Sinyor M, Levitt AJ, Cheung AH, Schaffer A, Kiss A, Dowlati Y et al. Does inclusion predictors of response to antidepressants in the GENDEP project. Pharmacoge- of a placebo arm inﬂuence response to active antidepressant treatment in nomics J 2009; 9: 225–233. randomized controlled trials? Results from pooled and meta-analyses. J Clin 122 Buckholtz JW, Meyer-Lindenberg A. MAOA and the neurogenetic architecture of Psychiatry 2010; 71:270–279. human aggression. Trends Neurosci 2008; 31:120–129. 129 Tedeschini E, Fava M. Goodness TM, Papakostas GI. Relationship between 123 Pavlov KA, Chistiakov DA, Chekhonin VP. Genetic determinants of aggression probability of receiving placebo and probability of prematurely discontinuing and impulsivity in humans. J Appl Genet 2012; 53:61–82. treatment in double-blind, randomized clinical trials for MDD: A meta-analysis. 124 Zubieta J-K, Heitzeg MM, Smith YR, Bueller JA, Xu K, Xu Y et al. COMT val158met Eur Neuropsychopharmacol 2010; 20:562–567. Genotype Affects μ-Opioid Neurotransmitter Responses to a Pain Stressor. Sci- 130 Stewart PA, Cameron T, Farb RI. Functional Neuroanatomy. 2012, Illustrated by ence 2003; 299: 1240–1243. Kevin Millar. Retrieved from http://fn.med.utoronto.ca/. 125 Brody AL, Mandelkern MA, Olmstead RE, Scheibal D, Hahn E, Shiraga S et al. 131 Cohen D, Consoli A, Bodeau N, Purper-Ouakil D, Deniau E, Guile J-M et al. Pre- Gene variants of brain dopamine pathways and smoking-induced dopamine dictors of placebo response in randomized controlled trials of psychotropic release in the ventral caudate/nucleus accumbens. Arch Gen Psychiatry 2006; 63: drugs for children and adolescents with internalizing disorders. J Child Adolesc 808–816. Psychopharmacol 2010; 20:39–47. 126 Jutkiewicz EM, Roques BP. Endogenous Opioids as Physiological Anti- 132 Khan A, Leventhal RM, Khan SR, Brown WA. Severity of depression and response depressants: Complementary Role of Delta Receptors and Dopamine. Neu- to antidepressants and placebo: an analysis of the Food and Drug Administra- ropsychopharmacology 2012; 37:303–304. tion database. J Clin Psychopharmacol 2002; 22:40–45. 127 Zhang H, Torregrossa MM, Jutkiewicz EM, Shi Y-G, Rice KC, Woods JH et al. 133 Stein DJ, Baldwin DS, Dolberg OT, Despiegel N, Bandelow B. Which factors Endogenous opioids upregulate brain-derived neurotrophic factor mRNA predict placebo response in anxiety disorders and major depression? An analysis through δ-andμ-opioid receptors independent of antidepressant-like effects. of placebo-controlled studies of escitalopram. J Clin Psychiatry 2006; 67: Eur J Neurosci 2006; 23: 984–994. 1741–1746.